The goal of this proposal is to build a more complete picture of ligand-protein recognition by examining free, intermediate and bound states in order to reveal the why and when of binding mechanisms and kinetic behavior. Non-covalent molecular recognition plays a crucial role in biology, chemistry and medicine. Exploring binding pathways and the transient intermediate states during binding will help elucidate mechanisms that include binding, allostery, induced fit, gated control associations, and the free energetics o binding, which will later guide molecular designs. Molecular simulations play an increasing role in studying binding thermodynamics and kinetics, important in both biology and chemistry. Used in combination with experiments, simulations integrate data and interpret experiments; then contribute to the design of novel molecules with preferred affinities and/or kinetic properties. Kinetic data also can be used as a critical differentiator and predictor for drug efficacy and safety. However, real molecular systems are quite complicated, and computational tools usually are either very time-consuming or over-simplify a biological system. Therefore, the major motivation is that we need projects to further develop new computational methods to both efficiently and accurately model molecular association pathways in order to understand the binding mechanisms, solvent effects and kinetic behavior. Guided by strong preliminary results and existing methods developed by our group, three specific aims are proposed: 1) Develop and apply multi-scale methods to model ligand-protein binding in order to understand binding mechanisms and kinetic behavior; 2) Investigate the role of waters and free energy landscape in binding processes; 3) Adapt and apply the new methods and integrate experiments to peptide- protein binding to study protein function and assist peptide design. The approach is innovative that it involves bringing new methodological breakthroughs to enable realistic modeling of binding pathways and expanding the classical view of molecular recognition. The proposed research is significant, because it provides computational tools for us to study binding processes, free energy surfaces and solvent effects, and reveals fundamental mechanisms of molecular association.